Humanity's Finest - Referrals

Note: At the moment, this opportunity is not available to applicants from California and New York.

—

We'll cut to the chase: we're looking for the world's best experts to take on the world's smartest models.

In Humanity’s Last Exam, we introduced the most challenging reasoning benchmark for frontier AI models. So far, the highest-performing system—OpenAI’s Deep Research—has achieved only 26% accuracy. We are collaborating with leading AI labs to identify the most effective data to improve AI reasoning capabilities in expert domains, and we aim to publish a new paper presenting our findings.

To do this, we are assembling a team of elite individuals who are the utmost experts in their respective fields. Our shared goal is to create PhD+ level problems that current state-of-the-art LLMs cannot correctly solve. This team will work collaboratively to produce datasets that will be available to our partner research groups, the world’s most advanced AI laboratories.

We're looking for unicorns to help us write some of the hardest problems intelligence has ever seen — do you think you can do this?

Why Join?

• Have the opportunity to co-author a research paper analyzing how effectively this data enhances model reasoning.
• Get access to an exclusive community of world-class researchers in a variety of domains; Each task submission will be open to peer review.
• Receive up to $540 for each contributed problem and solution pair

What’s Next?

• Still unsure? Watch this quick video of what comes after you hit "Apply Now" below
• If you're up for the task, apply below!